Introduction
Allergic rhinitis (AR) and asthma are chronic inflammatory disorders of the upper and lower airways that share immunologic and pathophysiologic mechanisms.1 In AR and allergic asthma, sensitization to aeroallergens such as pollen, house dust mites, animal dander, and moulds contributes to IgE-mediated airway inflammation.2,3 Traffic-related nitrogen oxides and fine particulate matter are additional environmental exposures associated with adverse respiratory health outcomes.4 AR typically presents with sneezing, nasal itching, rhinorrhea, and nasal obstruction, whereas asthma commonly presents with wheezing, cough, chest tightness, and breathlessness.2,3 They frequently coexist and can interact, increasing symptom burden and disease severity.5 Consistent with the “united airway, united disease” concept, the term combined allergic rhinitis and asthma syndrome (CARAS) has been used to describe concurrent AR and asthma.6 Reported estimates suggest that approximately 40% of patients with AR have asthma, whereas up to 80% of patients with asthma have AR.7,8 Comorbidity is associated with greater clinical burden and poorer health-related quality of life (HRQoL) than either condition alone.9
HRQoL captures the patient-perceived effects of disease and treatment on physical, emotional, and social functioning. In respiratory allergy, patient-reported outcome measures (PROMs) complement clinician-assessed symptoms, lung function, and other objective indicators.10 Incorporating patients’ values, preferences, and perceived burden is also consistent with contemporary evidence-based guideline development and shared clinical decision-making.11,12
Several disease-specific instruments are available for AR or asthma, including the Rhinoconjunctivitis Quality of Life Questionnaire (RQLQ) and Asthma Quality of Life Questionnaire (AQLQ).10 Because these instruments were developed for individual diseases, either questionnaire alone may not capture the combined upper- and lower-airway burden of CARAS, whereas administering both increases respondent burden. The Control of Allergic Rhinitis and Asthma Test (CARAT) assesses both conditions but focuses primarily on disease control rather than HRQoL.13
The RhinAsthma Patient Perspective (RAPP) is an 8-item PROM developed to assess the impact of coexisting rhinitis and asthma on daily life.14 Its brevity makes it suitable for routine care. The ARIA–EAACI Task Force has proposed the RAPP as a patient-centered digital biomarker for monitoring HRQoL in allergic respiratory disease, reinforcing its potential role in integrated rhinitis–asthma management.15 Versions have been validated in English, Polish, Portuguese, and Spanish populations,16–19 and the instrument is cited in the 2024 Chinese expert consensus on CARAS.6 However, a culturally adapted and psychometrically validated Chinese version is unavailable. We therefore translated and culturally adapted the RAPP into simplified Chinese and evaluated its measurement properties in Chinese adults with CARAS.
Materials and Methods
Translation and Cross-Cultural Adaptation
The 8-item RAPP was translated from Italian into simplified Chinese and culturally adapted in accordance with the International Society for Pharmacoeconomics and Outcomes Research Principles of Good Practice for translating and adapting PROMs.20 Written permission to translate and reproduce the instrument was obtained from the copyright holder.
Two independent native Chinese-speaking translators fluent in Italian produced forward translations. The translators, a respiratory physician, and an otolaryngologist reconciled the translations into a single Chinese version. Two independent bilingual translators who were blinded to the original instrument then back-translated this version into Italian.
An expert committee comprising translators, clinicians, and a psychometrician compared the source, forward-translated, and back-translated versions for semantic, idiomatic, experiential, and conceptual equivalence. Discrepancies were resolved by consensus to produce a prefinal Chinese version.
Cognitive debriefing was conducted with 20 adults with CARAS. Participants completed the prefinal questionnaire and were asked about item clarity, comprehensibility, cultural relevance, and response-option suitability. Patient feedback informed minor wording revisions, after which the expert committee approved the final Chinese RAPP.
An English version of the questionnaire is provided in Table S1 for reference.
Patients and Data Collection
Participants were recruited from three tertiary hospitals in Henan Province, China, between October 2024 and March 2025 using convenience sampling. Eligible participants were native Chinese speakers aged 18 years or older. CARAS was operationally defined as concurrent guideline-confirmed AR and asthma; participants in the isolated-disease groups met diagnostic criteria for either AR or asthma and had the other condition excluded.21,22 To evaluate responsiveness, participants with CARAS completed a follow-up assessment after a 4-week consensus-guided, syndrome-based traditional Chinese medicine (TCM) treatment period.
Clinicians at all three hospitals received standardized training in the syndrome differentiation criteria and therapeutic principles specified in the 2024 integrated traditional Chinese and Western medicine consensus for CARAS.6 Treatment followed the shared three-phase, six-pattern framework (Table S2). For each identified syndrome pattern, clinicians selected a formula recommended in the consensus, with individualized adjustments guided by the corresponding therapeutic principles.
Exclusion criteria were respiratory or otorhinolaryngologic diseases other than AR or asthma; pregnancy or lactation; critical illness; severe depression, anxiety, or another psychiatric or cognitive disorder that could interfere with participation; and visual, hearing, speech, or other impairment that precluded independent questionnaire completion.
Conventional recommendations suggest enrolling at least 5 to 10 participants per item for initial PROM validation.23,24 We targeted 180 participants with CARAS to support classical test theory (CTT) and item response theory (IRT) analyses, allowing for nonresponse or incomplete questionnaires. For secondary exploratory group comparisons and ROC analyses, we prespecified pragmatic recruitment targets of 80 participants with isolated AR and 80 with isolated asthma, without a separate formal sample size calculation. These group sizes fall within the range of 50–99 participants per group categorized as adequate for known-groups validity in the COSMIN Study Design checklist (2019).25 The study protocol was approved by the Ethics Committee of the First Affiliated Hospital of Henan University of Chinese Medicine, and all participants provided written informed consent before enrollment.
For test‑retest assessment, participants with CARAS who were in the clinical-control phase completed the RAPP again after 2 weeks. For longitudinal assessment, participants with CARAS repeated the RAPP and ACT after the 4-week treatment period.
Study Measures
The RAPP comprises 8 items rated from 1 (not at all) to 5 (very much) with a 2-week recall period. Item scores are summed to yield a total score of 8 to 40, with higher scores indicating greater disease impact and poorer HRQoL. At baseline, all participants completed the RAPP and provided demographic information.
Participants with CARAS also completed the RQLQ, AQLQ, and Asthma Control Test (ACT). These instruments were scored according to their respective manuals; higher RQLQ scores indicate poorer HRQoL, whereas higher AQLQ and ACT scores indicate better HRQoL and asthma control, respectively.
Reliability
Internal consistency was assessed using Cronbach α and item–total correlations. Test‑retest reliability was evaluated using an intraclass correlation coefficient (ICC) based on two RAPP assessments 2 weeks apart. Cronbach α and ICC values of 0.70 or higher were considered acceptable.24
Validity
Validity was evaluated across three domains: structural validity, convergent validity, and discriminant validity.24,26 Structural validity was assessed with item–total correlations and confirmatory factor analysis of the proposed single-factor model. Standardized factor loadings greater than 0.50 were considered satisfactory. Model fit was considered acceptable when the comparative fit index (CFI) and non-normed fit index (NNFI) were at least 0.90 and the root mean square error of approximation (RMSEA) and standardized root mean square residual (SRMR) were no greater than 0.08.24
Convergent validity was assessed using Pearson correlations between RAPP total scores and RQLQ, AQLQ, and ACT scores. Positive correlations with the RQLQ and negative correlations with the AQLQ and ACT were expected because of their opposing scoring directions. Absolute correlation coefficients greater than 0.40 were considered supportive.27
Discriminant validity was examined by comparing RAPP scores across CARAS clinical phases, ACT-defined control levels, and disease groups using one-way analysis of variance. ACT-defined control was categorized as well controlled (20–25), partly controlled (16–19), or uncontrolled (5–15). As an exploratory item-discrimination analysis, item scores were compared between participants in the upper and lower 27% of the RAPP total-score distribution using independent-samples t tests.
Exploratory receiver operating characteristic analyses quantified score discrimination between CARAS and each isolated-disease group; these analyses were not intended to evaluate the RAPP as a diagnostic instrument.
Responsiveness
Responsiveness was examined by comparing RAPP and ACT scores before and after the 4-week treatment period using paired t tests, effect size (ES), and standardized response mean (SRM). ES was calculated as the mean change divided by the baseline standard deviation, and SRM as the mean change divided by the standard deviation of the change score. Changes were oriented so that positive ES and SRM values represented clinical improvement.
Because these are distribution-based indices, the analysis was interpreted as preliminary evidence of the ability to detect change rather than an estimate of clinically important change.
Differential Item Functioning
Differential item functioning (DIF) by sex was assessed using ordinal logistic regression. The RAPP total score was used as the matching variable, the item score as the dependent variable, and sex as the grouping variable. To limit type I error, statistical significance was set at P <0.01; items with P ≥0.01 were considered free of statistically detectable DIF.28
Item Analysis Based on Item Response Theory
IRT was used to characterize item-level performance after dimensionality had been examined by factor-analytic methods. The discrimination parameter (a), four ordered difficulty thresholds (b1–b4), and item information were estimated for each item. Item performance was considered acceptable when a ranged from 0.30 to 3.00, the difficulty thresholds increased monotonically and ranged from −4.00 to 4.00, and average item information exceeded 5/n, where n is the number of items.29,30 Category response curves were inspected for monotonic lowest and highest response categories and distinct peaks for intermediate categories.
Feasibility
Feasibility was evaluated using the questionnaire return rate, proportion of returned questionnaires with complete item data, and completion time. A return and completion rate greater than 85% and a mean completion time of less than 5 minutes were considered acceptable.
Statistical Analysis
IRT analyses were performed using MULTILOG version 7.03, confirmatory factor analysis using LISREL version 8.80, and all other analyses using SPSS version 26.0. Continuous variables are summarized as mean ± standard deviation and categorical variables as number (percentage). Unless otherwise specified, tests were two-sided and P < 0.05 was considered statistically significant.
Questionnaires with at least 20% missing item data were classified as invalid; for questionnaires with less than 20% missing data, the prespecified approach was item-level mean imputation.
Results
The primary psychometric cohort comprised 164 participants with CARAS. An additional 80 participants with isolated AR and 80 with isolated asthma were included in exploratory group comparisons, yielding an overall analytic sample of 324 participants (148 men and 176 women; age range, 19–73 years). Patient characteristics are shown in Table 1.Table 1Baseline Characteristics of ParticipantsVariablesCategoryTotal Sample (N = 324)CARAS (n=164)AR (n=80)Asthma (n=80)Age, years44.96 ± 12.4346.70±12.8942.79±11.3643.58±12.12SexMale148 (45.7%)70 (42.7%)41 (51.3%)37 (46.3%)Female176 (54.3%)94 (57.3%)39 (48.8%)43 (53.8%)Marital statusUnmarried54 (16.7%)25 (15.2%)16 (20.0%)13 (16.3%)Married251 (77.5%)133 (81.1%)57 (71.3%)61 (76.3%)Widowed19 (5.9%)6 (3.7%)7 (8.8%)6 (7.5%)Educational levelElementary school and below50 (15.4%)29 (17.7%)9 (11.3%)12 (15.0%)Junior high school75 (23.1%)48 (29.3%)12 (15.0%)15 (18.8%)High school108 (33.3%)38 (23.2%)37 (46.3%)33 (41.3%)College and above91 (28.1%)49 (29.9%)22 (27.5%)20 (25.0%)OccupationFarmer62 (19.1%)30 (18.3%)15 (18.8%)17 (21.3%)Worker99 (30.6%)50 (30.5%)28 (35.0%)21 (26.3%)Skilled worker41 (12.7%)22 (13.4%)7 (8.8%)12 (15.0%)Staff54 (16.7%)29 (17.7%)14 (17.5%)11 (13.8%)Manager26 (8.0%)14 (8.5%)7 (8.8%)5 (6.3%)Retired20 (6.2%)10 (6.1%)4 (5.0%)6 (7.5%)Other22 (6.8%)9 (5.5%)5 (6.3%)8 (10.0%)Note: Data are presented as mean ± SD or n (%).Abbreviations: AR, allergic rhinitis; CARAS, combined allergic rhinitis and asthma syndrome; SD, standard deviation.
Reliability
Cronbach α for the RAPP total score was 0.876. Deleting any individual item did not improve internal consistency (Table 2). Fifty-seven participants in the clinical-control phase completed the RAPP again after 2 weeks.
The total-score ICC was 0.835 (P <0.001). Item-level ICCs exceeded 0.70 except for item 4 (ICC = 0.628) and item 8 (ICC = 0.697) (Table 3). RAPP total scores at the two assessments were positively associated (Figure 1).Table 2Cronbach α if an Item Was DeletedItemCronbach Alpha if Item DeletedItem 1 Stuffy or runny nose, sneezing, or itchy nose0.853Item 2 Itchy eyes, watery eyes, sore eyes, and eyes redness0.856Item 3 Difficulty in concentrating0.875Item 4 Wheezing, cough, chest tightness, shortness of breath0.848Item 5 Problems in sleeping (ie, night awakenings)0.853Item 6 Having to avoid certain areas and environments0.858Item 7 Having to take drugs0.874Item 8 Limitations in performing some activities (ie, working, studying, sports)0.863Note: Cronbach alpha for the total RAPP score was 0.876.Table 3Test-Retest Reliability of the RAPPItemFirst AssessmentSecond AssessmentICCItem 11.60 ± 0.701.63 ± 0.700.783*Item 21.65 ± 0.721.61 ± 0.650.739*Item 31.77 ± 0.951.67 ± 0.690.703*Item 41.67 ± 0.741.58 ± 0.650.628*Item 51.60 ± 0.651.54 ± 0.660.723*Item 61.81 ± 0.741.67 ± 0.690.704*Item 72.00 ± 0.911.96 ± 0.890.823*Item 81.98 ± 0.791.93 ± 0.780.697*Total score14.07 ± 3.3313.60 ± 3.060.835*Notes: Data are presented as mean ± SD; *P<0.001.Abbreviations: ICC, intraclass correlation coefficient; RAPP, RhinAsthma Patient Perspective; SD, standard deviation.Figure 1Scatterplot of RAPP total scores at baseline and week 2.A scatter plot showing RAPP total scores at baseline versus week 2, from 5 to 30 and 0 to 25.Notes: The fitted regression line is y = 0.8229x + 2.018, where x and y represent RAPP total scores at baseline and week 2, respectively.Abbreviation: RAPP, RhinAsthma Patient Perspective.
Validity
Structural Validity
All item–total correlations were statistically significant and exceeded 0.70 except those for item 3 (r = 0.630) and item 7 (r = 0.610) (Table 4). Confirmatory factor analysis supported the proposed single-factor structure: CFI = 0.983, NNFI = 0.976, RMSEA = 0.053, and SRMR = 0.036. Standardized factor loadings ranged from 0.52 to 0.93 (Figure 2).Table 4Item-Total Correlations of the RAPPItemCorrelation with Total ScoreP valueItem 10.790< 0.001Item 20.766< 0.001Item 30.630< 0.001Item 40.823< 0.001Item 50.781< 0.001Item 60.748< 0.001Item 70.610< 0.001Item 80.703< 0.001Abbreviation: RAPP, RhinAsthma Patient Perspective.Figure 2Confirmatory factor analysis model of the simplified Chinese RAPP.A path diagram of a single-factor RAPP model linking RAPP to eight items with loadings.Abbreviation: RAPP, RhinAsthma Patient Perspective.
Convergent Validity
RAPP scores correlated positively with RQLQ scores (r = 0.613) and negatively with AQLQ (r = −0.659) and ACT scores (r = −0.772; all P <0.001). The directions were consistent with instrument scoring: higher RAPP and RQLQ scores indicate greater impairment, whereas higher AQLQ and ACT scores indicate better health status or asthma control.
Discriminant Validity
In the exploratory item-discrimination analysis, all 8 item scores differed between the upper and lower 27% of the RAPP total-score distribution (Table S3). Analyses of discriminant validity showed differences in RAPP scores across CARAS clinical phases (F = 88.765, P <0.001), ACT-defined control levels (F = 107.761, P <0.001), and disease groups (F = 16.646, P <0.001) (Table 5). Exploratory receiver operating characteristic analysis yielded an area under the curve of 0.680 (95% CI, 0.610–0.750) for CARAS versus isolated AR and 0.665 (95% CI, 0.592–0.738) for CARAS versus isolated asthma (Figure 3).Table 5Discriminant Validity of the RAPP Across Clinical and Disease GroupsSubgroupRAPP Score (Mean ± SD)FP valueDisease phaseAcute exacerbation (n=52)25.52 ± 4.1088.765< 0.001Chronic persistent (n=55)19.91 ± 5.70Clinical control (n=57)14.07 ± 3.33ACT-defined control levelUncontrolled, ACT 5–15 (n=38)26.87 ± 4.29107.761< 0.001Partly controlled, ACT 16–19 (n=74)20.20 ± 4.99Well controlled, ACT 20–25 (n=52)13.62 ± 2.82Disease groupCARAS (n=164)19.66 ± 6.4616.646< 0.001AR alone (n=80)15.60 ± 3.86Asthma alone (n=80)15.95 ± 4.44Notes: Higher ACT scores indicate better asthma control; higher RAPP scores indicate greater disease impact.Abbreviations: ACT, Asthma Control Test; AR, allergic rhinitis; CARAS, combined allergic rhinitis and asthma syndrome; RAPP, RhinAsthma Patient Perspective; SD, standard deviation.Figure 3Exploratory receiver operating characteristic curves of RAPP scores for discriminating CARAS from isolated AR and isolated asthma.Two line graphs showing receiver operating characteristic curves for RAPP scores across two comparisons.Notes: (A) CARAS versus isolated asthma: AUC = 0.665 (95% CI: 0.592–0.738).
(B) CARAS versus isolated AR: AUC = 0.680 (95% CI: 0.610–0.750). The diagonal lines indicate chance-level discrimination.Abbreviations: AR, allergic rhinitis; AUC, area under the curve; CARAS, combined allergic rhinitis and asthma syndrome; CI, confidence interval; RAPP, RhinAsthma Patient Perspective.
Responsiveness
RAPP scores decreased and ACT scores increased after the 4-week treatment period (both P <0.001), indicating improvement in the expected directions. For the RAPP, the ES was 0.41 and the SRM was 0.79; corresponding ACT values were 0.42 and 0.52 (Table 6).Table 6Responsiveness of the RAPP and ACT After 4 WeeksInstrumentBaseline4 WeeksESSRMtP valueRAPP19.66 ± 6.4616.98 ± 5.310.410.79−10.116< 0.001ACT18.13 ± 2.9119.35 ± 2.410.420.526.691< 0.001Notes: Data are presented as mean ± SD. ES was calculated as mean change divided by baseline SD.
SRM was calculated as mean change divided by the SD of the change score.Abbreviations: ACT, Asthma Control Test; ES, effect size; RAPP, RhinAsthma Patient Perspective; SD, standard deviation; SRM, standardized response mean.
Differential Item Functioning
At the prespecified threshold of P <0.01, ordinal logistic regression detected no statistically significant sex-related DIF for any RAPP item (Table S4).
IRT-Based Item Analysis
Factor-analytic diagnostics supported essential unidimensionality, with a Kaiser–Meyer–Olkin value of 0.905 and a statistically significant Bartlett test of sphericity (P <0.001). The scree plot (Figure S1) and principal component analysis identified one component with an eigenvalue greater than 1, explaining 54.13% of the variance; the ratio of the first to second eigenvalue was 5.64.
IRT discrimination parameters ranged from 1.15 to 2.86, and difficulty thresholds ranged from −1.67 to 3.49; all values were within the prespecified ranges (Table S5). Item 7 had the highest upper difficulty threshold and was the only item with average information below the prespecified value of 0.625 (Table S6). Category response curves showed the expected monotonic patterns for the lowest and highest categories and distinct peaks for intermediate categories (Figure 4).
Figure 4Item characteristic curves for the 8 RAPP items based on item response theory analysis.A) Line graph with x-axis label, Latent HRQoL impairment (theta), unit not shown, ranging from negative 3 to 3. Y-axis label, Probability, unit not shown, ranging from 0.0 to 1.0. Five curves are plotted.
Curve 1 decreases from near 1.0 at x equals negative 3 to near 0.0 by x equals 1 to 3. Curve 2 is bell-shaped with a peak near x equals negative 0.5 at about y equals 0.55. Curve 3 is bell-shaped with a peak near x equals 0.8 at about y equals 0.50.
Curve 4 is bell-shaped with a peak near x equals 1.6 at about y equals 0.33. Curve 5 increases from near 0.0 at x equals negative 3 to near 1.0 by x equals 3. B) Same axes as A.
Five curves are plotted. Curve 1 decreases from near 1.0 at x equals negative 3 to near 0.0 by x equals 1 to 3. Curve 2 peak is near x equals negative 0.3 at about y equals 0.50.
Curve 3 peak is near x equals 0.9 at about y equals 0.55. Curve 4 peak is near x equals 1.8 at about y equals 0.40. Curve 5 increases from near 0.0 at x equals negative 3 to near 0.9 by x equals 3.
C) Same axes as A. Five curves are plotted. Curve 1 decreases from about y equals 0.9 at x equals negative 3 to near 0.0 by x equals 3.
Curve 2 is broader with a peak near x equals 0 at about y equals 0.45. Curve 3 is broader with a peak near x equals 1.2 at about y equals 0.33. Curve 4 is broader with a peak near x equals 2.0 at about y equals 0.30.
Curve 5 increases from near 0.0 at x equals negative 3 to about y equals 0.6 by x equals 3. D) Same axes as A. Five curves are plotted.
Curve 1 decreases from near 1.0 at x equals negative 3 to near 0.0 by about x equals 0 to 1. Curve 2 peak is near x equals negative 0.2 at about y equals 0.65. Curve 3 peak is near x equals 0.8 to 1 at about y equals 0.60.
Curve 4 peak is near x equals 2 at about y equals 0.75. Curve 5 increases from near 0.0 at x equals negative 3 to above y equals 0.8 by x equals 3. E) Same axes as A.
Five curves are plotted. Curve 1 decreases from near 1.0 at x equals negative 3 to near 0.0 by x equals 1 to 3. Curve 2 peak is near x equals negative 0.4 at about y equals 0.55.
Curve 3 peak is near x equals 0.8 at about y equals 0.48. Curve 4 peak is near x equals 2.0 at about y equals 0.50. Curve 5 increases from near 0.0 at x equals negative 3 to about y equals 0.8 by x equals 3.
F) Same axes as A. Five curves are plotted. Curve 1 decreases from near 1.0 at x equals negative 3 to near 0.0 by x equals 3.
Curve 2 peak is near x equals negative 0.6 at about y equals 0.45. Curve 3 peak is near x equals 0.8 at about y equals 0.55. Curve 4 peak is near x equals 2.2 at about y equals 0.55.
Curve 5 increases from near 0.0 at x equals negative 3 to about y equals 0.6 to 0.7 by x equals 3. G) Same axes as A. Five curves are plotted.
Curve 1 decreases from about y equals 0.8 to 0.9 at x equals negative 3 to near 0.0 by x equals 3. Curve 2 peak is near x equals negative 1 at about y equals 0.45. Curve 3 peak is near x equals 1 at about y equals 0.50.
Curve 4 peak is near x equals 2.6 at about y equals 0.45. Curve 5 increases from near 0.0 at x equals negative 3 to about y equals 0.4 by x equals 3. H) Same axes as A.
Five curves are plotted. Curve 1 decreases from about y equals 0.9 at x equals negative 3 to near 0.0 by x equals 3. Curve 2 peak is near x equals negative 1 at about y equals 0.35 to 0.4.
Curve 3 peak is near x equals 0.6 at about y equals 0.65. Curve 4 peak is near x equals 2 at about y equals 0.40 to 0.45. Curve 5 increases from near 0.0 at x equals negative 3 to about y equals 0.6 to 0.7 by x equals 3.A set of line graphs showing item haracteristic curves for the 8 items of the RAPP.Notes: (A–H) correspond to Items 1–8, respectively.
Response categories 1 through 5 are shown in black, blue, magenta, green, and cyan, respectively. θ represents latent HRQoL impairment, with higher values indicating greater impairment.Abbreviations: HRQoL, health-related quality of life; RAPP, RhinAsthma Patient Perspective.
Feasibility
Of 180 RAPP questionnaires distributed to participants with CARAS, 164 were returned and eligible for analysis, corresponding to a return rate of 91.1%. All returned questionnaires had complete item data; therefore, no imputation was required. Mean completion time was 1.89 ± 1.56 minutes.
Discussion
Routine CARAS assessment often emphasizes clinician-rated symptoms, physical findings, and objective tests, potentially underestimating the impact perceived by patients in daily life.31 A brief patient-reported outcome measure can therefore add a complementary patient-centered dimension to clinical evaluation. The RAPP was derived from the 30-item RHINASTHMA questionnaire, whose length and complexity limited routine use.32 By retaining 8 clinically relevant domains—nasal and ocular symptoms, sleep disturbance, concentration, lower-airway symptoms, environmental avoidance, medication burden, and activity limitation—the RAPP reduces respondent burden while preserving the multidimensional experience of CARAS. In this study, its Chinese version demonstrated acceptable reliability, validity, feasibility, and preliminary responsiveness, supporting further use and evaluation in clinical and research settings.
The Chinese RAPP showed strong internal consistency (Cronbach α = 0.876), comparable to the original Italian version (α = 0.867)14 and the English, Polish, and Portuguese versions (α = 0.87, 0.85, and 0.82, respectively).16–18 This cross-language consistency suggests that the 8 items capture a coherent underlying construct while remaining sufficiently concise for routine administration.
The test-retest ICC of 0.835 met the prespecified acceptability threshold but was numerically lower than the 0.97 reported for the English and Portuguese versions.16,18 This finding requires caution because clinical assessment and phase classification were used to select the 57 retested participants, without a patient-reported global rating of change (GRC) anchor to verify stability over the 2-week interval. Clinical control does not necessarily imply absence of patient-perceived daily burden, as medication burden and environmental avoidance may persist despite controlled symptoms. For test-retest reliability, the relevant requirement is stability of this burden between assessments.
Unrecognized changes in patient-perceived burden may therefore have contributed to score variation alongside measurement error. ICC values also depend on between-participant variability, which limits direct comparisons across studies. The available data cannot establish the reasons for the lower ICC, and confirmation in a larger sample with GRC-based stability assessment is warranted.
Structural validity was supported by the unidimensional factor solution, which explained 54.13% of the variance and was comparable to findings for the original and other language versions.14,16–19 All standardized loadings exceeded 0.50, indicating that each item contributed meaningfully to the common construct. The comparatively lower item–total correlations for concentration difficulty (item 3) and medication burden (item 7) may reflect influences beyond airway symptoms, including sleep, treatment behavior, and patient expectations. Moderate-to-strong correlations with the RQLQ, AQLQ, and ACT in the expected directions further supported convergent validity: higher RAPP scores were associated with poorer HRQoL and asthma control.
Collectively, these findings support the conceptual coherence of a single patient-perceived burden spanning upper- and lower-airway disease.
Discriminant validity was supported across several clinically relevant comparisons. Item scores differentiated the upper and lower 27% of the total-score distribution, and total scores followed the expected gradient across ACT-defined control levels and CARAS clinical phases. The mean score in the clinical-control phase approximated the previously reported threshold for optimal HRQoL,33 supporting clinical interpretability.
Higher scores in CARAS than in isolated AR or asthma were also consistent with the greater burden of combined upper- and lower-airway disease.34,35 The modest ROC areas, however, reinforce that the RAPP should quantify patient-perceived burden and monitor HRQoL rather than diagnose or distinguish disease entities.
Responsiveness indicates whether a patient-reported outcome measure can detect change over time.36 Using the ACT as a parallel clinical reference, we evaluated RAPP change within a consensus-guided, three-phase, six-pattern TCM framework.6 Treatment principles were strictly standardized across the three hospitals through an expert-consensus matrix, with herbal prescriptions individualized within these predefined principles. After 4 weeks, the RAPP showed a small ES (0.41) and a moderate SRM (0.79) according to conventional criteria,37 with improvement also observed in ACT scores. Heterogeneity in prescriptions and clinical responses may affect both the mean change and its variability.
ES standardizes mean change by the baseline SD, whereas SRM uses the SD of individual change scores; for a given mean change, greater change-score variability reduces SRM. The contributions of pharmacological effects, behavioral or contextual influences, and natural symptom fluctuation could not be separated. These distribution-based findings therefore provide preliminary, context-specific evidence of short-term responsiveness, without establishing the efficacy of a specific formula or clinically meaningful change.
Future multicenter longitudinal studies with longer follow-up should estimate the minimal clinically important difference (MCID) using anchor-based methods.36 Patient global ratings of change (GRC) in the combined impact of rhinitis and asthma could serve as the primary anchor, with ACT change as a complementary anchor. After confirming that the anchors correlate adequately with RAPP change scores, candidate thresholds could be estimated from the mean RAPP change in minimally improved patients and ROC analyses distinguishing minimally improved from stable patients. These estimates should be evaluated alongside measurement error and validated in independent samples.
The addition of item response theory extends previous RAPP validation studies based mainly on classical test theory.38 Most items showed acceptable discrimination and ordered difficulty thresholds, indicating useful performance across levels of HRQoL impairment. Item 7 provided relatively little information and required a higher level of latent impairment for endorsement of the highest category, possibly reflecting heterogeneity in treatment intensity, perceived dependence, and medication burden.39 No statistically significant sex-related DIF was detected, providing preliminary support for comparable item performance in men and women; confirmation in larger subgroups remains desirable.
Strengths of this study include systematic cross-cultural adaptation; evaluation of complementary classical and item response theory-based measurement properties; assessment of feasibility and short-term responsiveness; and direct collaboration with the instrument developers. The brief completion time and high response rate further support applicability in routine clinical and research settings.
Several limitations warrant consideration. Convenience sampling from three tertiary hospitals in one Chinese province may limit geographic and clinical generalizability. Although the sample of 164 participants with CARAS was adequate for CTT analyses, it was modest for polytomous IRT parameter estimation and subgroup DIF analyses.
This may limit the precision and stability of item discrimination and category threshold estimates. The IRT findings should therefore be interpreted cautiously and confirmed in larger independent samples. Clinical stability during the 2-week retest interval was inferred from clinical phase rather than verified using a patient-reported GRC.
Future studies should use a GRC anchor to identify participants reporting no change in the combined impact of rhinitis and asthma over the retest interval and assess test-retest reliability in this stable subgroup. Responsiveness was evaluated over only 4 weeks within a specific treatment framework. Floor and ceiling effects, measurement error, broader measurement invariance, and the minimal clinically important difference were not established.
Larger multicenter longitudinal studies should address these properties.
Conclusions
The Chinese version of the RAPP demonstrated acceptable reliability, validity, and feasibility, with preliminary evidence of responsiveness, for assessing HRQoL in Chinese adults with CARAS. Its brevity and favorable psychometric performance support its potential use in routine clinical assessment, follow-up monitoring, and research. Multicenter studies should confirm longitudinal measurement properties and establish the minimal clinically important difference.